[kumarUjjawal]

Description

This PR fixes VACUUM INTO so it preserves hidden rowid values for ordinary rowid tables instead of renumbering them in the destination database.

Motivation and context

Issue #5805 reported that VACUUM INTO produced a destination DB where rowids were renumbered (e.g. rowid 5 became 1), while SQLite preserves rowid values for ordinary rowid tables.

Closes #5805

Description of AI Usage

[turso-bot]

Please review @avinassh

[avinassh]

why do we need this

[kumarUjjawal]

thanks for the feedback, I removed current_table_column_names and switched alias detection to existing schema metadata.

[avinassh]

can you also add sqltest

you could also add a compatibility test in tests/integration/query_processing/test_vacuum.rs showing how sqlite preserves and turso did not

[Pavan-Nambi]

could u please add test cases for below scenarios too

explicit INTEGER PRIMARY KEY
and vaccum into with indexes(parital,normal,unique)

just to be more safe as they have specific tests in sqlite too.

[Copilot]

Pull request overview

This PR updates Limbo’s VACUUM INTO implementation to preserve hidden rowid values for ordinary rowid tables (matching SQLite behavior), and adds regression/compatibility coverage to prevent future rowid renumbering regressions.

Changes:

• Preserve hidden rowid values during VACUUM INTO table-copy by explicitly selecting/inserting an available rowid alias (rowid, _rowid_, oid) when not shadowed by a real column.
• Add integration tests covering rowid preservation (including shadowed rowid aliases) and compatibility comparisons against SQLite’s VACUUM INTO.
• Add runner-level SQL tests for VACUUM INTO error cases (empty path, disallowed in transaction).

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
core/vdbe/execute.rs	Adjusts the VACUUM table-copy SELECT/INSERT generation to include an accessible rowid alias so destination tables preserve rowid values.
tests/integration/query_processing/test_vacuum.rs	Adds integration/compat tests validating rowid preservation and behavior when rowid aliases are shadowed by user columns.
testing/runner/tests/vacuum_into.sqltest	Adds SQLTest coverage for VACUUM INTO argument/transaction error handling.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codspeed]

Merging this PR will improve performance by 17.71%

⚡ 1 improved benchmark
✅ 278 untouched benchmarks
⏩ 105 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	concat_blobs	1.7 µs	1.5 µs	+17.71%

---

_{Comparing kumarUjjawal:fix/vaccum_into_rejection (40f77de) with main (0ce4e6a)}

Footnotes

1. 105 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

These new tests build the VACUUM INTO SQL by interpolating dest_path directly inside single quotes. This will break if the temp path contains a ' character (common on Windows when usernames contain apostrophes) and can make the test suite environment-dependent. Consider reusing escape_sqlite_string_literal() for the Turso VACUUM INTO path as well (or otherwise ensure proper SQL string literal escaping).

[Copilot]

data_columns.remove(idx) is guarded by turso_assert!, which will panic in non-antithesis builds if schema/PRAGMA column ordering ever diverges (e.g., from corruption or future PRAGMA changes). Since VACUUM INTO can be run on user-provided DBs, prefer returning a LimboError::Corrupt/ParseError when the index is out of bounds instead of panicking.

Suggested change

	turso_assert!(
	idx < data_columns.len(),
	"rowid alias column index out of bounds for table columns",
	{ "idx": idx, "columns_len": data_columns.len() }
	);
	data_columns.remove(idx);
	excluded_rowid_alias_column = true;
	if idx < data_columns.len() {
	data_columns.remove(idx);
	excluded_rowid_alias_column = true;
	}